DOI: 10.1002/ijc.70760 ISSN: 0020-7136

Human Protein Biomarkers Outperform Bacterial Proteins in Colorectal Cancer Detection

Roza C. M. Opperman, Sofie Bosch, Alex A. Henneman, Thang V. Pham, Sander R. Piersma, Tim G. J. de Meij, Meike de Wit, Beatriz Carvalho, Remond J. A. Fijneman, Gerrit A. Meijer, Evelien Dekker, Connie R. Jimenez, Nanne K. H. de Boer

ABSTRACT

Colorectal cancer (CRC) generally develops from precancerous adenomatous lesions, including advanced adenomas (AA), through a sequence of carcinogenic molecular alterations. The composition of the gut microbiota and the fecal proteome is known to differ between individuals with AA and CRC compared to controls, providing opportunities for biomarker discovery. This study aims to assess the potential of bacterial proteins as non‐invasive biomarkers alone or in combination with human proteins. Therefore, two fecal sample sets from a colonoscopy‐controlled population were analyzed: sample set 1 (CRC: 12, AA: 10, controls: 20) and sample set 2 (CRC: 79, AA: 40, controls: 129). Mass spectrometry‐based proteomics was used to identify and quantify proteins in the fecal samples of the two series. Overlapping differential proteins ( p  < 0.05) were used to establish biomarker panels through logistic regression and leave‐one‐out cross‐validation. Three biomarker panels were evaluated: one containing bacterial proteins, one combining human and bacterial proteins, and one containing human proteins alone. In total, 5400 overlapping proteins were identified across both sample sets. For combined CRC and advanced adenoma detection, the human‐bacterial and human‐only panels achieved the best performance (AUCs 0.88–0.90; sensitivities 56%–68%), exceeding human hemoglobin alone (AUCs 0.78–0.86; sensitivity 49%) and the bacterial panel (AUCs 0.67–0.75; sensitivities 17%–46%). In conclusion, human protein biomarker panels demonstrate superior diagnostic performance over bacterial proteins, particularly for CRC detection, indicating that the diagnostic value of fecal metaproteomics in this setting is mainly driven by host‐derived proteins.